While Bev Harris of BlackBoxVoting.org may or may not be doing wonderful things on the topic of electronic voting with no paper trail, she is pushing a bad methodology in the analysis of Florida voting results.
In her article on:
http://www.democraticunderground.com/discuss/duboard.php?az=view_all&address=104x2636130
(more below the fold)
She writes:
2. CRUNCH THE NUMBERS: Discrepancies please, and hurry. E-mail them to tips@blackboxvoting.org . Pass the word. Need source documents, too. ASAP. Follow your nose, or join the Black Box Scavenger Hunt: Pick a county. Look at small counties, as we are seeing many discrepancies in those. Look in any state. Get the official number of registered voters, Dem and Republican. Get the number of votes cast on Nov. 2, Republican and Dem. Make a grid like this, filling in the right numbers:
# reg. voters % # votes cast %
Rep 100 33% 150 50%
Dem 200 67% 150 50%
Totl 300 100% 300 100%
We can find out a lot from this procedure, very hard data, that will make a real difference. As soon as you have finished a county, e-mail it to us. Do as many counties as you can.
Unfortunately, although these numbers can be crunched, they have NO VALIDITY for deterimining whether there are errors in the vote count, they merely LOOK alrming in many cases.
To have some idea of what the numbers should look like for this election, you would need to do a much more thorough analysis, checking to see what they looked like in previous elections, what the demographics of the county look like, and how those demographics have voted both this year nationally and in past elections in that region.
People often cross party lines when voting. This tends to be a regional phenomina often down to the county level, and it tends to vary by year. This cannot be ignored, but that it exactly what her methodology does.
She does herself and her cause no favors by pushing this crap.
Apparently, my mission today is to keep pointing out her bad methodology on dKos while my unit tests churn away.
(edit) Gustavo: Bad Typer. Cannot even spell "Bev" properly in the headline the first time out.